ICT Market Maker Models: How MMBM and MMSM Work

Written by Dominic Walsh · Published 23 Jul 2026 · Last updated 17 Aug 2026

ICT market maker models describe the full arc of a price delivery. Price builds a base, leaves it in an engineered move, posts a smart money reversal, and then returns back through the same levels. Michael Huddleston, the Inner Circle Trader (ICT), teaches two mirrored templates — the market maker buy model (MMBM) and the market maker sell model (MMSM), known together as MMXM. So by the end of this deep dive, you will be able to label every stage of the curve and locate where your entry belongs on it.

First, we define the model and the curve concept, using a real gold sell model with exact levels. Then we walk both templates stage by stage. Also covered: where the models sit inside the wider ICT workflow, and the honest limitations.

What Are ICT Market Maker Models?

A market maker model is a map of how an engineered price swing completes itself. Price leaves a range, runs one way to collect liquidity, and reverses. Then it retraces the whole trip, often reacting at the very levels it made on the way out. So if the method is new to you, start with what ICT trading is before working through the curve.

The chart below shows a market maker sell model on gold (XAUUSD, 1-hour) from mid-July 2026, with the consolidation, the buy-side legs, and the smart money reversal marked.

Trace the labels on that chart. Gold consolidated between 4,017 and 4,040 around July 14, building orders on both sides of a 23-dollar range. Manipulation followed: price ran up to 4,074, paused, and pushed once more to 4,081, sweeping the highs above the range. There the smart money reversal formed. Distribution then unwound the whole advance, carried price back through the range, and completed near 3,990 — a 91-dollar journey from the extreme to the finish.

Think of the model as a curve — a valley for the buy model, a hill for the sell model. Plainly, one side of the curve is the setup and the other is the payoff. Crucially, the two sides mirror each other. Each shelf built on the way out tends to get a revisit on the way back. Hence the model’s practical value — the mirror turns finished structure into forward targets.

The Anatomy of the Curve

Original consolidation

Every model starts with balance. Price moves sideways, building orders above and below. This range is the original consolidation, and it anchors the whole structure. Also, its midpoint often acts as the last magnet when the curve completes. On the gold chart, that anchor was the 4,017-4,040 box.

The manipulation legs

Next, price displaces away from balance. In a buy model this is the sell-side of the curve: one, two, or sometimes three legs lower, each pausing at a small shelf before it continues. These pauses matter. Indeed, they act as redistribution on the way down. Price then tends to re-trade them as flipped support or resistance on the way back up. Gold’s sell model showed the same thing inverted — the pause at 4,074 became a shelf on the buy-side of the hill.

Smart money reversal

Eventually price reaches its target: a higher-timeframe array or a pool of resting liquidity. There the smart money reversal (SMR) forms — the turn where the engineered flow gets absorbed and the real position builds. Then the evidence stacks: a sweep of an old extreme, divergence between correlated pairs, and displacement back in the opposite direction. On gold, the 4,081 push above the prior high was that terminal sweep.

The return through the curve

Finally, the payoff side. Price travels back through each small shelf from the manipulation phase, treating them as staging points. Completion arrives when price returns to the original consolidation. Meanwhile, extended deliveries continue into the liquidity beyond it — exactly how gold pressed on to 3,990, below the old 4,017 range floor.

The Market Maker Buy Model (MMBM)

Here is the market maker buy model as an ordered sequence. Also, every stage has a job — skipping one usually means the label is wrong.

1. Original consolidation. A sideways range forms on the 1-hour chart or higher. Both sides of the order book build up.
2. Sell-side of the curve. Price displaces lower in one to three legs, pausing at small shelves along the way. Meanwhile, sell-side liquidity accumulates under every new low.
3. Terminal raid. The final leg sweeps a meaningful old low or fills a higher-timeframe discount array — the reason the whole decline existed.
4. Smart money reversal. Rejection appears, then displacement up through a near-term swing point. Now the bias flips from selling rallies to buying dips.
5. Buy-side of the curve. Price re-trades each earlier shelf, now a re-accumulation zone, and mirrors the decline leg for leg.
6. Completion. The move returns to the original consolidation and, in strong deliveries, runs the buy stops resting above it.

Entries live on the right side of the curve. Plainly, buying inside stage two fights the manipulation, while buying after stage four trades with the reversal and against trapped sellers. Impatience here is the classic tuition payment: the trader who buys the second leg down because the chart “looks cheap” funds the raid that stage three still owes the market.

The Market Maker Sell Model (MMSM): The Mirror

The market maker sell model flips every element. Price bases, then rallies through one to three legs on the buy-side of the curve. Next it sweeps a major old high or taps a premium array, and the smart money reversal points down. Then the sell-side of the curve unwinds the advance, re-trading each pause as redistribution until price returns to the original consolidation and the sell stops below it. The gold chart at the top of this guide is exactly this template: a 4,017-4,040 base, a two-step rally to 4,081, and distribution back to 3,990.

Put trade logic on those gold numbers and the sell model becomes concrete. The short belonged after the reversal at 4,081 confirmed — displacement back below the 4,074 shelf — with the stop above the sweep high. From there the curve supplied the targets in order: the shelf, the range ceiling at 4,040, the midpoint near 4,028, and finally the stops under 4,017 that broke toward 3,990. A trader needed no prediction, only the discipline to let the left side of the hill finish printing first.

Nothing new appears in the sell model. Indeed, that is the point: learn one template deeply and you own both, because MMXM is a single symmetric structure viewed from either side. Traders who struggle usually study the buy model in isolation and freeze when the mirror image appears. So drill both directions on replay data until the labels come automatically.

Whose Orders Drive the Curve

Strip away the labels and the curve is a story about positioning. During the consolidation, both breakout camps set traps for themselves: buy stops accumulate above the range, sell stops below. The manipulation legs then spend one side of that fuel. In a sell model, each push higher fills institutional selling against breakout buying, which is why the legs stall at shelves instead of trending away cleanly.

The reversal flips the pressure. Once the final sweep fills the remaining size, trapped buyers above the range hold losers, and heavy inventory sits sold at premium prices. Their capitulation supplies the momentum for distribution, while each old shelf offers a place to re-sell any bounce. No conspiracy is required — clustered stops and obvious levels produce the pattern on their own, in any market deep enough to attract both crowds.

This reading also explains why the mirror works. The shelves on the manipulation side are prices where real transactions occurred — inventory changed hands there. When price returns, the trader who sold the shelf on the way up defends the same area on the way down, and the trapped buyer who held through the reversal finally exits into it. Old business creates future interest. Hence shelves behave like appointments the market tends to keep, even though nothing forces attendance.

Reading the Curve in Live Conditions

Live markets never label their stages, so the workflow supplies the context. First, read the higher timeframe: the daily chart says which pool of liquidity price is drawing toward — the essence of ICT daily bias. So a buy model only deserves attention when that draw points up from a discount; a sell model needs the opposite.

Levels and Sessions

Stage recognition then leans on levels. Indeed, the shelves, gaps, and blocks that price respects on both sides of the curve are the ICT PD arrays. Premium arrays cap the sell model’s rallies; discount arrays floor the buy model’s declines. Meanwhile, the intraday rhythm follows sessions: manipulation legs cluster in the London kill zone (2:00-5:00 New York time), and reversals often complete during the New York morning. Our free forex market hours tool shows those session windows in your local time.

Nesting Lower Timeframes

Zoomed in, each day inside the curve follows the accumulation, manipulation, and distribution rhythm. The full guide to the ICT power of three covers that daily template and how it nests inside larger models. Pair the framing chart with a faster trigger chart as well: curves mapped on the 4-hour or 1-hour chart resolve their entries on the 15- or 5-minute chart, where the reversal shows its structure shift first. Charting all of this by hand takes screen time. So some traders let our MMXM ICT indicator mark ranges, curve legs, and reversal zones for you, alongside the broader set of ICT indicators for MT4 and MT5.

A Curve Across the Week

Stretch the rhythm over several days and a familiar weekly shape appears. Monday and early Tuesday often build or extend the consolidation. Then the manipulation legs tend to run on Tuesday and Wednesday, with each London session pushing the curve deeper and each New York afternoon stalling it at a shelf. The terminal sweep frequently lands midweek, and the return side unwinds into Thursday and Friday. Treat this as a tendency, not a timetable — data weeks and trending regimes reshuffle it freely. Still, knowing the typical schedule stops you from expecting a full reversal an hour after the base breaks.

A EURUSD Buy Model Walkthrough, Entry to Target

Suppose EURUSD bases between 1.0860 and 1.0880 for two days on the 1-hour chart. Then price breaks down in two legs. First it drops to 1.0838 and pauses at a tight shelf near 1.0840; later it falls to 1.0804, sweeping a prior daily low at 1.0810 during the London kill zone.

Rejection follows at 1.0804, and displacement carries price up through 1.0824 — the smart money reversal. Now the entry sequence begins. On the 15-minute chart, the pullback after that displacement retests 1.0826, and a small structure shift turns it higher. The long fills at 1.0826 with the stop at 1.0801, below the raid low, for 25 pips of risk.

Targets come from the curve itself. The 1.0840 shelf marks the first checkpoint, the range floor at 1.0860 the second, and the stops above 1.0880 the completion objective — 54 pips from entry, a little over 2R. Each reaction on the way up happened at a level the decline created. Hence the value of marking the shelves in real time, before the turn.

Manage the position stage by stage rather than all at once. A partial at the 1.0840 shelf pays for the trade, the stop moves to entry at the range floor, and the final piece hunts the stops above 1.0880. If price stalls under a shelf for many candles without progress, respect the message — half-finished curves are common, and banking the structural targets you did reach beats donating them back while waiting for completion.

Common Mistakes with Market Maker Models

Six habits sink most curve traders. Each has a direct correction.

1. Counting legs with certainty. Nobody knows the count in real time. Correction: treat every leg as possibly non-final until the reversal confirms.
2. Buying arrival, not reversal. Touching a discount array is not a signal. Correction: wait for the sweep plus displacement through a swing point.
3. Labeling every range a consolidation. A base matters only when a higher-timeframe draw sits beyond it. Correction: name the draw first.
4. Studying one side only. Buy-model specialists freeze in sell conditions. Correction: drill both templates on replay until they feel symmetric.
5. Holding through the whole curve blind. Full-curve trades need staged partials. Correction: scale out at shelves and the range midpoint.
6. Forcing models onto news trends. Data-driven trends often skip the mirror leg. Correction: stand down around top-tier releases.

Notice the shared root: every mistake substitutes imagination for confirmation. The curve invites storytelling because it explains so much in hindsight, and the cure is procedural — demand the sweep, demand the shift, and let the checklist below veto anything the narrative invented. A model that survives those vetoes is worth risk; one that needs your generosity is not.

Curve Trading Checklist

Run the list before committing to a stage-four entry.

1. Higher-timeframe draw on liquidity identified.
2. Original consolidation marked, midpoint noted.
3. Manipulation legs and shelves labeled as they form.
4. Terminal leg sweeps an obvious old extreme.
5. Displacement through a swing point confirms the reversal.
6. Entry taken at the first re-traded shelf, not mid-air.
7. Stop beyond the raid extreme, size computed from it.
8. Targets laddered: shelves, range, then the stops beyond it.

The list doubles as a grading tool. Eight yes answers describe the textbook delivery worth full planned risk; six or seven describe a setup worth reduced size; fewer means the chart is not printing a model, whatever the eye wants to see. Write the answers down before entry — memory is a generous editor after the fact.

Where Market Maker Models Fail

The obvious weakness is hindsight bias. Curves look clean on finished charts, yet in real time the leg count is unknowable — what reads as a terminal raid can become leg two of three. So trading stage four requires evidence of reversal, not just arrival at a level.

Other limits deserve respect. Leg counting is subjective, and two competent traders can label the same decline differently. Also, trending weeks can skip the return trip entirely, leaving the mirror half-finished. News shocks rewrite targets mid-curve, and models built without a higher-timeframe draw are decoration, not analysis. Thus MMXM works best as a framing device that organizes bias, entries, and targets — the model orders your decisions, it does not make them for you.

Keep the statistics honest as well. No fixed completion percentage exists for the curve, and none of the numbers floating around social media survive contact with a real journal. What you can measure is your own base rate: of the models you label in real time, how many confirm a reversal, and how many of those reach the range. Track that over a quarter, per pair and per session, and you will know precisely how much weight your labeling deserves.

A Failed Reversal, Priced Out

Here is the classic trap in numbers. EURUSD bases between 1.1440 and 1.1460, then breaks down in two legs to 1.1408 and 1.1381, sweeping an old low at 1.1385. The arrival looks terminal. Yet the bounce stalls at 1.1402 without ever closing above the nearest swing point at 1.1408 — no displacement, no shift, no confirmed reversal.

A third leg then breaks 1.1381 and runs to 1.1352, and what looked like stage four becomes leg three of the decline. The protective rule is structural: no structure shift, no entry — arrival alone earns nothing. Traders who bought the 1.1385 sweep early paid for the lesson, while the stop under the raid low contained the cost for anyone who at least defined risk. Log the miss afterward: the leg count you assumed, the quality of the bounce, and where the real higher-timeframe draw sat. Most failed models were answering a draw one level deeper than the trader marked.

Related Concepts

Two siblings deepen the model. The terminal raid at the curve extreme is, in miniature, the turtle soup strategy — a fade of a raided extreme with the stop beyond the wick. Meanwhile, buy-side and sell-side liquidity explains the fuel the curve consumes on each side. Read both and the model stops feeling like a diagram; it starts reading like order flow with a schedule. Then revisit the gold and EURUSD walkthroughs above with those two lenses applied — the raids, pools, and shelves resolve into a single connected story rather than separate labels.

FAQ

What does MMXM mean?

MMXM is shorthand for “market maker X model,” where X stands for either buy or sell. So it covers both the market maker buy model (MMBM) and the market maker sell model (MMSM), since the two are mirror images of one structure.

What is the smart money reversal?

The smart money reversal is the turn at the far end of the curve, where the manipulation legs end and the real move begins. Indeed, it tends to form after a liquidity sweep at a higher-timeframe array, confirmed by displacement back through a swing point. On the gold chart above, it printed at 4,081.

How many legs does the manipulation side have?

ICT charts commonly show one to three legs, each separated by a small consolidation. Still, you only know the count for sure after the reversal confirms, which is why entries belong on the return side of the curve.

Are ICT market maker models the same as Wyckoff schematics?

They rhyme without being identical. Wyckoff’s accumulation and distribution schematics describe the same base-run-reverse behavior with older vocabulary, developed a century earlier. ICT adds session timing, liquidity pools, and PD arrays. Traders comfortable with one framework usually pick up the other quickly.

Do market maker models imply someone controls the market?

No single desk steers a market as deep as EURUSD or gold. Instead, the model shows how engineered liquidity runs and dealer inventory flows tend to shape price around obvious levels — a behavior map, not a conspiracy claim.

Can beginners trade the full curve?

Beginners do better trading one stage — usually the reversal plus the first re-accumulation — rather than holding through the whole curve. Truly managing a position across every leg takes practice and screen time. Results are not guaranteed; past performance is not indicative of future results.